C300 MkII ISO Colour Shifts

[Mathew Williams]

Hi, I'm noticing quite a colour shift when changing ISO. Even when quite a small change - going from ISO 1000 to ISO 1250 there is quite a shift from a red bias to a green bias - as shown in this link -

C300II ISO Colour Shifts between ISO 1000 and 1250 on Vimeo

I've also noticed some strange colour banding, as shown in the attached image.

Anyone else have similar, or any idea what's causing these issues? Thanks,

Matt

[Barry Goyette]

Hi Matt,

It would be great if you could give us more information concerning the footage. What gamma setting were you using? Also is the footage graded in anyway?

First culprits would be ABB....did you Auto Black Balance before shooting this? The banding you are seeing would appear to be CMOS smear caused by the highly overexposed area and usually seen in the darkest shadow tones. (the darkest shadows are also the place where you usually see color shifts in one direction or the other on this camera, ABB can help to clean that up.) Without know what you did to the footage or what gamma you used, it's hard to tell, but if I had to guess, the color switching between ISO changes is not that unusual in this case. This footage is extremely limited in it's range....all we see are areas that are over and underexposed in the same shot and we are looking at an area that, once graded, should be rendered as black. You may be testing the limits of the cameras range here. How bout a fill card? :-)

[Mathew Williams]

Hi Barry, thanks for the reply. The settings were Clog3, production camera, BT709. The video clip hasn't been graded at all (but I did scale in to 130% just to crop out the person in shot to protect their privacy). This was an event shoot so no lighting used, or time to be using bounce panels etc. I was just interested to see this colour shift and so thought I'd ask about it. I actually noticed the same thing on a shoot yesterday as well. I did an ABB a few days ago, maybe that changed something - I will try doing another though not sure why that should change anything?

The still image has had the midtones raised in post just to make the smear stand out more clearly - thanks for the explanation it is somehow caused by the area of overexposure - are you able to explain a bit more about why this happens, and why it shows up in the dark area?

Thanks again.

[Barry Goyette]

Hi Matt,

So...in general, you want to do an ABB every time you fire up the camera, typically wait for 20 mins or so for it to warm up and then do it. I also find it's helpful to take a hard look at your shadows after ABB to make sure things are looking neutral. If not, ABB again. The shadow end of a log gamma is looking at the most fragile, noisy portion of the sensor readout, and it will exhibit color and noise changes based on changes in temperature and also image processing, (which changing the gain or "ISO" is) and so doing the ABB is going to at least balance the image before you get started.

Now...typically shooting in log gamma you are going want to try to "shoot to the right" a little so that your image is light, and shadow areas can be pushed darker in post. Raising mid-tones in LOG is generally the opposite of what you're supposed to be doing.

CMOS smear is caused by overexposed areas (smarter minds than me can explain the electronic aspect of this), but is visible in the darkest shadows that are adjacent to or in line with those overexposed areas...again if your workflow is correct, rarely is it seen in graded footage as it's in the area you are generally crushing to black or near black. Cmos smear is visible to some degree on almost every camera with a cmos sensor, although the C300markII seems to exhibit it more in Log2 than other cameras.

Here's where it gets interesting...your first problem...the color balance changes in the shadows when you change ISO...welll that may just be CMOS smear caused by upping the gain on an already saturated highlight. We're not seeing the telltale line there because the highlight runs the full hight of the frame.

Just so you know how common cmos smear is...when you look at a DSC Labs Xyla dynamic range chart you'll notice a peculiar wedge shape to the graduated steps of the chart. The purpose for this wedge shape was to avoid misreadings due to CMOS smear, as the chart is designed so the first step is always clipped. That step is smaller in height compared to the darker steps, so any smear affects only the center of each of the darker steps allowing for a proper reading of the margins to ascertain DR of a camera.

[Mathew Williams]

Thanks Barry, that's very informative and good advice regards the ABB and waiting for the camera to warm up before performing it. I will try this and see how I get on. Thanks!